Reader Feedback on Database Modernization Still Unknown Territory
August 18, 2008 Hey, Ted
I seem to have missed something on the DDS versus DDL topic, which Dan Burger wrote about in Database Modernization Still Unknown Territory. Explain why and how DDL is faster if they both run on the same box. The speed of the disk drives is the same. If there is some “super algorithm” for DDL, why can’t it be applied to DDS? Is there some design feature of DDS that just makes it slow compared to DDL, or is DDL really just something new just because IBM thought it was time for a change, kinda like the i versus AS/400 thing?
There is more to I/O than hardware. Input and output are governed by software. This is nothing new. Buffering has been around for years.
I refer you to Chapter 2 of the Redbook Modernizing IBM eServer iSeries Application Data Access–a Roadmap Cornerstone. You will see from the diagrams that native I/O and the Classic Query Engine (CQE) are implemented above the Machine Interface, but the SQL Query Engine (SQE) is implemented below the MI. So, no, the performance is not the same.
Could the “super algorithms” be applied to DDS? Of course some of them could, but IBM has chosen not to apply them to DDS. SQL is the strategy. SQL gets the development dollars.
Did IBM change directions from DDS to SQL just for the sake of change? No, by no means. IBM realizes that SQL can do what people need to be done, and DDS cannot. And that is the reason I use SQL.